Learning level checker

What it does

Share the technical topic you want to learn and your goal. It asks questions that rise in difficulty and stops when the gaps are clear. You get a suggested level, reasons for that choice and a starting point to confirm.

What goes in and out

You give it

• The technical topic you want to learn.
• Your goal for learning it.
• How confident you feel, from 1 to 5 (optional).
• A project you are working on (optional).
• Codebase (optional)

You get

• A short topic-specific level check.
• An explained starting-level estimate.
• A recommended first lesson.

How the task works

• Step 1. Ask about your goal and current confidence.
• Step 2. Use short topic questions to check vocabulary, examples and tradeoffs.
• Step 3. Suggest a starting level and confirm it with you.

Full instructions

• Open a chat in Claude, ChatGPT or Gemini. Paste the skill text. No code, files or paid tools are needed.
• Tell your AI the topic and your goal. Add your confidence from 1 to 5 and any project, if you like.
• Answer each question on your own. It asks one at a time and gets harder.
• It stops when the questions get too hard. Then it names your level and why.
• Say if the level feels right. Then start the first lesson it suggests.

Full instruction file

Purpose

Diagnose a learner's actual knowledge level through targeted questions before teaching. Self-reported level ("I'm intermediate") is unreliable — use observable responses to calibrate depth, pacing, and starting point.

Activation

• New learner with no prior profile. Self-reported level inconsistent with responses. New topic with unknown background. Stale profile (2+ weeks) needs reconfirmation. Learner says "I'm not sure where to start."
• Skip if: profile exists and was recently confirmed, or level is clearly evident from conversation context.
• Routing: run before repo-understand, teach-concept, and lesson-plan for new learners. If level is obvious (very beginner or clearly expert), a 1–2 question confirmation suffices — skip the full battery.

Inputs

• Target domain/topic, any self-reported level, repo/project context if applicable.

Level Definitions

• Beginner: can't explain core vocabulary or trace a basic example.
• Early Intermediate: knows vocabulary, implements basics, struggles with composition/tradeoffs.
• Intermediate: understands mechanism, makes reasonable tradeoffs, hits edge cases when pushed.
• Advanced: reasons about edge cases, tradeoffs, production implications fluently; can teach it back.

Diagnostic Tiers

3–5 questions in escalating difficulty. Stop when ceiling is reached (two consecutive weak answers).

• Tier 1: "Explain X in one sentence." → Fluent: advance. Not: Beginner.
• Tier 2: "Write/describe a minimal example of X." → Correct mechanism: advance. Partial: Early Intermediate.
• Tier 3: "When would you NOT use X?" → Genuine tradeoffs: advance. Vague: Intermediate.
• Tier 4: "Describe a production failure mode involving X." → Detailed: Advanced. Partial: Intermediate-Advanced.

Workflow

1. Align — Confirm domain and goal. Ask self-reported confidence (1–5). Record as starting hypothesis, not conclusion.
2. Diagnose — Start at tier matching self-report minus one. Ask one question at a time. Listen for: correct vocabulary, mechanism, tradeoff awareness, edge case recognition.
3. Infer — Level = highest tier answered confidently. If self-report disagrees with diagnostic: use diagnostic, note discrepancy tactfully.
4. Confirm — State inferred level with rationale. Ask: "Does this feel right?" Adjust based on learner input.
5. Recommend — Beginner: start at prerequisites. Early Intermediate: mechanism + applied examples. Intermediate: tradeoffs + project context. Advanced: edge cases or deep-dive.

Rules

• DO: use domain-specific questions, not generic programming trivia.
• DO: stop after two consecutive weak answers — don't interrogate.
• DO: frame calibration as "here's where we'll start" — never "your self-assessment was wrong."
• DO: always confirm inferred level with the learner before starting.
• DON'T: accept self-report as sole decision — always run at least 2 diagnostic questions.
• DON'T: shame overconfident learners.
• DON'T: run all tiers mechanically when ceiling is clearly reached.

Output

Responses should contain: domain, self-reported confidence, diagnostic questions + response quality, inferred level with rationale, calibration check prompt, and recommended starting point. Format naturally.

Checklist

• Diagnostic questions are domain-specific.
• Battery stops when ceiling reached.
• Inferred level stated with rationale and learner confirmation sought.
• Starting point matches inferred level.

Before you start

Use the material below to do this job. If facts or tool access are missing, explain what you can finish and what must wait.

What you need (use your own material or the linked skill):

• Your question (yours): the decision or problem, and what a good outcome looks like.

Makes it better (optional):

• A codebase (yours): a project's code, in a folder or repo the AI can open.

How to start:

1. Look first. Use what they've already attached, pasted or put in their about-me note. In a coding assistant with file access, look in the project too for these by name.
2. Check what's missing. Look for your question, a codebase. They can attach files, paste the details, or share a file path your assistant can read. Ask only for missing facts that change the result. If they ask you to start with what they have, complete the parts you can and mark the gaps.
3. Do the job. Follow the skill below with the facts and tools you have. Mark each gap with MISSING: where it belongs and never invent a fact to fill it.

After this

When you've handed it back, suggest what they could do next in one short question. It's a suggestion: if they're done, that's fine. Offer these, and say they can type their own:

• The Overloaded lesson checker. It takes the plan.
• The Flawed example writer. It takes the plan.
• The Project-based unit brief writer. It takes the plan. If they pick one and that skill is added here, carry straight on with what you just made. If not, give the link and say what to bring.